OAI/NumberTheory/JointDickman/Arithmetic/MovingRoughExpansion.lean

GitHub source file

4 declarations, 0 contain sorry, 0 declared as axioms

The moving-cutoff expansion with explicit errors

The classical expansion, Taylor approximation, and correction-moment tails are now combined for the actual rough squarefree summatory function. No moving-cutoff asymptotic is used as an input.

  1. def OAI.JointDickman.powerExpansionError line 19
  2. theorem OAI.JointDickman.powerExpansionError_le line 28

    Source documentation

    For the nonpositive powers occurring here, the moment-tail sum is bounded by a single fixed logarithmic power.

    theorem powerExpansionError_le {A C a L : ℝ} {H P V : ℕ}
        (hA : 0 ≤ A) (ha : a ≤ 0) (hL : 1 ≤ L)
        (hlog : 1 ≤ Real.log P) (hV : 0 < V) :
        powerExpansionError A C a H P V L ≤
          C * L ^ (a - (H + 1)) * (Real.log P) ^ (Real.exp 1 + (H + 1 : ℕ)) +
            A * (∑ k ∈ range (H + 1), |rpowShiftCoeff a k| * (k.factorial : ℝ) * 2 ^ k) *
              (Real.log P) ^ (Real.exp 1 + H) / (V : ℝ) ^ ((1 / Real.log P) / 2) := by
    …

    Mentions: OAI.JointDickman.powerExpansionError, OAI.JointDickman.rpowShiftCoeff

  3. theorem OAI.JointDickman.correction_sdPolynomial_reindex line 72
  4. theorem OAI.JointDickman.moving_rough_expansion_explicit line 91

    Source documentation

    The actual expansion with separately identified classical, Rankin, Taylor, and moment-tail errors. Its coefficients are the defined moments.

    theorem moving_rough_expansion_explicit
        (hSD : PublishedInputs.SquarefreeSelbergDelangeInput)
        (hM : PublishedInputs.PrimeReciprocalMertensInput) {z : ℝ}
        (hz : z = 1 / 4 ∨ z = 1 / 2) :
        ∃ c : ℕ → ℝ, c 0 = squarefreeLeadingConstant z ∧ 0 < c 0 ∧
          ∃ A : ℝ, 0 < A ∧ ∀ H : ℕ, ∃ Csd : ℝ, ∃ C : ℕ → ℝ,
            0 ≤ Csd ∧ (∀ j, 0 ≤ C j) ∧ ∀ (P V : ℕ) (Y : ℝ),
              2 ≤ P → 1 ≤ Real.log P → 0 < V → 9 ≤ Y → (V : ℝ) ^ 2 ≤ Y →
              |roughSquarefreeSummatory (Nat.primesLE P) z Y -
                Y * ∑ ν ∈ range (H + 1), roughCoefficient c (Nat.primesLE P) z ν *
                  (Real.log Y) ^ (z - 1 - ν)| ≤
                Csd * Y * (Real.log Y / 2) ^ (z - 2 - H) *
    …

    Mentions: OAI.JointDickman.PublishedInputs.SquarefreeSelbergDelangeInput, OAI.JointDickman.PublishedInputs.PrimeReciprocalMertensInput, OAI.JointDickman.squarefreeLeadingConstant, OAI.JointDickman.roughSquarefreeSummatory, OAI.JointDickman.roughCoefficient